A hypothesis class is a family of hypotheses with the same function structure and typically needs to be pre$-$defined before learning.

A hypothesis class is a family of hypotheses with the same function structure and typically needs to be pre$-$defined before learning.

True

False

Question at position $5$

$5$

$1$ point

Question at position $5$

The process to determine the parameters in a model is called

The process to determine the parameters in a model is called

Optimization

Feature extraction

Inference

Prediction

Question at position $6$

$6$

$1$ point

Question at position $6$

Empirical risk minimization is a PAC learning algorithm, not an agnostic PAC learning algorithm.

Empirical risk minimization is a PAC learning algorithm, not an agnostic PAC learning algorithm.

True

False

Question at position $7$

$7$

$1$ point

Question at position $7$

Choose the correct statement.

Choose the correct statement.

A richer hypothesis class will have smaller approximation error

To get an optimal trade$-$off between estimation error and approximation error, it requires choosing a good optimization algorithm, especially to adjust the approximation error

A richer hypothesis class will have smaller estimation error

We should always choose a richer hypothesis class for machine learning, because it provides much better fitting capacity to simulate the behavior of the underlying concept function.